Autonomous navigation of tethered drone strings in cluttered environments through real-time obstacle detection and geometric tether control

This paper aims to fill the gap by identifying and critiquing the fundamental limitation of current tethered-drone systems by introducing a detailed framework of autonomous navigation in a highly obstacle-filled setting. The concept of the serially connected quadrotor is further stretched to allow real time collision avoidance and intact tether and system stability in GPS-denied cluttered environments. The suggested solution would combine vision-based sensing at the drones to detect obstacles and map the environment at a 3D scale. The hierarchical control structure is an integration of three components wherein geometric tether attitude control is involved, multi-objective model predictive control (MPC) path planning, and adaptive tether slack management in the case of emergency maneuvers. Three new contributions are made. A dynamic tether reconfiguration algorithm is first computed, which is an algorithm that calculates intermediate waypoints that remodel the tether geometry around obstacles without compromising tautness and stability. Second, the distributed artificial potential field technique, in which inter-drone communication is performed, eliminates drone-obstacle and tether-obstacle collision. Third, adaptive emergency protocol will allow an ability to avoid quickly using controlled temporary slack with assured system recovery. The MPC formulation is a combination of trajectory tracking, obstacle clearance, tether tension maintenance and control effort which subjects safety constraints to thrust limit, tension limit and clearance distance minimum. Simulated experiments show that 10 tethered drones can navigate successfully in a complex obstacle field (0.8m average clearance and safe tension range (3-8N)). In the proposed method, the risk of collision is reduced by 73%, and the chances of mission are improved by 42%, based on an average MPC solve time of 50.8 ms per control cycle and a deadline-aware fallback mechanism, which ensures that all control output are produced at every sampling instant with a bounded-latency. The given framework allows tethered drone delivery in the limited industrial areas, search and rescue use in cities, and infrastructure inspection as well as GPS-deprived areas that need autonomous obstacle avoidance functions.